DIY Reactive Variable Attenuator (Demo and Schematic inside)

[SnickSound]

EDIT: Please do not build this :) It was a flawed design from the beginning, by putting the reactive part in parallel with the speaker I forgot to account for series resistance amplifying the "reactance" of the speaker. This is why the response was all over the place and I tried to contain it by having extra switches to change inductance values. Either way... it's not a good design. I'll be leaving the schematic up either way for anyone curious and to better understand what is being discussed in this thread, but please don't attempt this unless you have some breakthrough ideas.

Note that JohnH on the Marshall forum has a simpler yet better idea: by having an inductor load both in parallel and series with the speaker, the sound is maintain and the amp still sees a proper load. The only thing is that scaling this to multiple attenuation levels would be quite complex so it still uses resistive l-pads to go quieter. But since the grunt of the signal is absorb by the initial reactive load and the series resistance interact with the speaker reactance, response is maintained both from the speaker and the amp's point of view. It is lacking the resonance component though, but the series resistance's interaction with the speaker will still result in a signal boost around there, it just won't reflect back to the amp as much so "resonance" controls might not work as intended.
Here's the thread: http://www.marshallforum.com/threads/simple-attenuators-design-and-testing.98285/
(It's my next project)


Thought some of you might find this interesting, since most reactive attenuator are either fixed attenuation (e.g. Two-Notes Captor only has -20dB) or very expensive (e.g. UA The Ox, Iron Man II), and some of those don't even have the cone resonance reactance (at least one of the 3 I mentioned doesn't) I decided to do my own research and experimentation.

Starting from the Randall Aiken paper on a reactive load box, I combined that with 3 attenuation level (-6, -12 and -18dB, selectively sending more and more current through the reactive side instead of the speaker) plus an L-Pad for fine volume control.

I still have some tweaking to do, the "presence" part of the circuit is either too much or too little right now, and I'm losing more low-mids than I'd like, but it's getting pretty close.

This is an actual recording of the amp (a 15W Plexi type amp) at various attenuation levels, down to "apartment level", using a Greenback loaded 2x12 and a Sennheiser e906 (reamped with a Radial ProRMP for consistency), both in solo and in a band context:

And this is the schematic of the device as of this recording:

https://1.bp.blogspot.com/-nDN1GJCfH8Y/WqBCXzD9lYI/AAAAAAAAEO0/cgxlpZS2N18GmzTmFdrux5fUt3jIzrZvwCLcBGAs/s1600/Reactenuator%2BFirst%2BWorking%2BPrototype%2Bv0.6%2B%2B%2528Fixed%2529.gif

[PRR]

Nice.

Questions about drawing:

L-pad is a 3-leg device but yours only has 2?

The second -6dB switch seems to have lost a link in drawing or posting?

My own curiosity: I believe (after re-drawing for my comprehension) that there is a setting for S1 S2 which shorts-out all the mechano-acoustic elements of the fake speaker seen from node "S". Thinking more, I realize this may indeed be intended?

As I read it, the -6dB switches could be SPDT for the same function. Yes, the switch cost may not change. But one less jumper to go missing.

L1 L3- two 0.5mH may be a dime cheaper than two 1mH? They don't strictly need to be "16AWG" if the DCR is accounted against R1-R3 (unless DCR is larger than most crossover coils).

What does C2 bring to the party? I realize it represents a discrete thing in the physical speaker. However at this level it is just parallel to a 220uFd, and another 68nFd is like a mite on a mouse. (Mouse and cone have similar mass; 68nF||220uF is 220uF for all practical purpose, especially with the tolerances on cone-mass and e-cap uFd.) Yes, C2 "bypasses C1's ESR and ESL", but a modern e-cap has really low ESR/ESL and 68nFd isn't 8r until 300KHz, so "infinite" in the audio band.
___________________________

Nothing to do with your plan: your fake-speaker resonates at 94Hz. This is in the typical range of guitar speakers. But no specific rig is "typical". Suppose I bring an Altec 417, 69Hz. JBL E130 is 40Hz. Mod 5-30 is 125Hz. As I turn from direct to heavy-cut the "bass resonance" shifts 4 semitones up or 14 semitones down. In the extreme case, the big soft long-bass E-130 becomes a stiff-cone compact.

Such fake speakers "should have" a variable L2. This is not simple. Lemme think while it snows here.

[SnickSound]

Nice.

Questions about drawing:

L-pad is a 3-leg device but yours only has 2?

The second -6dB switch seems to have lost a link in drawing or posting?

Sorry about that, thought I had uploaded the corrected schematic. It's fixed now.

My own curiosity: I believe (after re-drawing for my comprehension) that there is a setting for S1 S2 which shorts-out all the mechano-acoustic elements of the fake speaker seen from node "S". Thinking more, I realize this may indeed be intended?

That's correct. There isn't really a use for it (except if using this as a resistive load for testing and such), but since it's just a prototype I wanted the option of comparing reactive vs resistive.
Note that the "shorted" resonance option was changed to have a 8 ohm resistor instead of just a wire, this is simply because the big inductor has non-trivial DCR and R3 was added to compensate for that, if I bypass L2 completely the total impedance would be lower than intended.

As I read it, the -6dB switches could be SPDT for the same function. Yes, the switch cost may not change. But one less jumper to go missing.

Lol, you're totally right! Sometimes our minds tend to overcomplicate things!

L1 L3- two 0.5mH may be a dime cheaper than two 1mH? They don't strictly need to be "16AWG" if the DCR is accounted against R1-R3 (unless DCR is larger than most crossover coils).

That's what the local surplus store had in stock for fairly cheap, they were out of 0.47mH. Parts availability played a factor, I like to build with stuff I can easily source locally.

What does C2 bring to the party? I realize it represents a discrete thing in the physical speaker. However at this level it is just parallel to a 220uFd, and another 68nFd is like a mite on a mouse. (Mouse and cone have similar mass; 68nF||220uF is 220uF for all practical purpose, especially with the tolerances on cone-mass and e-cap uFd.) Yes, C2 "bypasses C1's ESR and ESL", but a modern e-cap has really low ESR/ESL and 68nFd isn't 8r until 300KHz, so "infinite" in the audio band.

I tested with and without C2, top end sounds clearer with it. The big cap is from Solen (a bipolar electro, not one of their huge film caps), I figured it would be good quality, but the difference is noticeable.

___________________________

Nothing to do with your plan: your fake-speaker resonates at 94Hz. This is in the typical range of guitar speakers. But no specific rig is "typical". Suppose I bring an Altec 417, 69Hz. JBL E130 is 40Hz. Mod 5-30 is 125Hz. As I turn from direct to heavy-cut the "bass resonance" shifts 4 semitones up or 14 semitones down. In the extreme case, the big soft long-bass E-130 becomes a stiff-cone compact.

Such fake speakers "should have" a variable L2. This is not simple. Lemme think while it snows here.

I actually can go from 13mH to 18mH right now (second 5.1mH inductor bypassable with a switch) which takes it down to around 80Hz, but it sounds better at 94Hz so I didn't bother recording both options.

Thanks for the help!

[Skydragon]

Very interesting, thanks for sharing this, much appreciated. Please continue to share further as you refine the design.

fwiw I've just been doing some testing with a two notes torpedo captor and a clone deluxe reverb amp. Surprisingly, I found the -20db fixed attenuation on the captor too much. Even at 'bedroom' volumes, I was at 6-7 on the amp volume before I could get a usable speaker volume using the captor.

Despite really wanting to like the Captor, I also found the tone not to be great, too fizzy and not a lot like the tone I can get using VVR on the same amp at low volumes. Perhaps I was expecting too much from it (I'm sure as a pure dummy load for recording with IR's it works well, but I wanted a decent attenuator as well).
 
When you've finished your design work, I may have a go at building your attenuator, to work with a Tweed Deluxe I have.

[SnickSound]

Very interesting, thanks for sharing this, much appreciated. Please continue to share further as you refine the design.

fwiw I've just been doing some testing with a two notes torpedo captor and a clone deluxe reverb amp. Surprisingly, I found the -20db fixed attenuation on the captor too much. Even at 'bedroom' volumes, I was at 6-7 on the amp volume before I could get a usable speaker volume using the captor.

Despite really wanting to like the Captor, I also found the tone not to be great, too fizzy and not a lot like the tone I can get using VVR on the same amp at low volumes. Perhaps I was expecting too much from it (I'm sure as a pure dummy load for recording with IR's it works well, but I wanted a decent attenuator as well).
 
When you've finished your design work, I may have a go at building your attenuator, to work with a Tweed Deluxe I have.

At -20dB, your Deluxe Reverb is down to less than 1/4 of a watt. It's a very limited appllication, but at least it's there (unlike other dummy loads with no speaker attenuation options at all). I can see this being useful when recording at home, allowing me to monitor through a real cabinet at 0 latency.

From what I was able to find online, Two Notes loads don't have the "cone resonance" part of the equation, just the coil inductance. Still, lots of users are satisfied with their product, maybe there is some compensation going on when using the line output to feed speaker simulators (a sharp boost at around 80-100Hz would probably do). That would explain why they're excellent products for running through IRs, but not so popular for speaker attenuation. The resonance circuit requires a big inductor (the one I'm using is 6" in diameter and cost me 50$CAD, but there are other options), which tends to have enough DCR that you have to compensate for it (R3 on my schematic). I've found that it adds noticeable warmth to the sound though, making it less fizzy.

That said, I do want mine to also double as a pure load for feeding IRs. My live rig includes a Line6 HX Effects which has an IR loader, I want to run the amp into the "Reactenuator" then into the HX so that I can run some effects post amp and then send that to FoH and my IEM feed. I could go 100% silent, or keep a cabinet on stage to help with feedback and such, with the device allowing me to control how loud I want this cabinet to be.

One of my amps has an AB763 front end (like a blackface Deluxe Reverb), except with a Paraphase inverter and cathode biased 6973 power tubes (it started life as a Supro clone), I'll try and do some clips with that one when I get a chance so you can get a better feel for whether this design works for that type of amp.

[Skydragon]

Thanks that would be great (a clip of AB763 amp)

Btw - have you had any issues with the attenuator causing any interference with your guitar or amp? (I’m guessing it generates quite a bit of a magnetic field when being driven hard)

[SnickSound]

Thanks that would be great (a clip of AB763 amp)

Btw - have you had any issues with the attenuator causing any interference with your guitar or amp? (I’m guessing it generates quite a bit of a magnetic field when being driven hard)

I've had more issues with it being sensible to interference. First time I brought it to a gig, we had a WiFi mixer next to the amp and it causes all kinds of weird noises.
Also, if I put it too close to the input of the amp I get oscillations with higher gain.

It has to be put in a shielded box and kept away from the amp's input.

[silverfox]

Curious. Could someone run the circuit through one of those simulator programs and see the response characteristics?

silverfox.

[Johnh]

Curious. Could someone run the circuit through one of those simulator programs and see the response characteristics?

silverfox.

Ive been doing that modelling, and Ill post soon.

I think its very tricky to mix a fixed reactive load box with speaker output reduced by passive resistive steps, since the resistive load damps the reactive effects. But Im curious to see how close it can get, and theory problems are not real if it still sounds good and is safe.

The active attenuators have a 'purer' theory, sampling signal from a reactive load with high impedance so not to change it.

[Johnh]

Here is a circuit analysis, made with 5Spice (free download). I hope I have represented the circuit fairly. Let me know if not.





It represents panning the pot in snicksounds schematic, with the two switches set to max volume and tone switches off.


The schematic uses a model of a speaker response, which seems reasonable and I have used it three times. It is used once to represent the attenuator itself, and also two times to represent a real speaker connected to the attenuator, and also a non-attenuated speaker direct to the amp.


The amp is just a signal source, but it is very important in this analysis to capture a reasonable output impedance for the amp. Ive tested this on a couple of my Marshalls and in a tube guitar amp, it is quite high. I use 20 Ohms here. Its actually the key to how the natural reactance of the speaker develops a boost to highs and lows.


The top trace shows this, and represents an unattenuated speaker. The lower traces show the attenuated traces, sweeping the pot. The shape of the response is still there, but reduced in height at low volume. Hence you would expect some loss of highs and lows in the attenuated sound.   


I also tried with an extra 8 Ohms in series with the speaker and this helps to restore the highs.

Id love to hear the video, but its not available now.

[zutph3n]

Hi Guys,

This looks very interesting!!! Any progress on this?

thx,
zut

[pompeiisneaks]

Odd, not sure but I used to be subscribed to his channel and watched most of his vids but it's gone now.  Either he deleted it and all content or YouTube did for some reason?

His last time active on the forums was in May too, so he seems to now be AWOL.  :(

~Phil

[synchu]

Worry not. The entire YouTube is down now. So the video will be back (hopefully)...


Niki

[pompeiisneaks]

Worry not. The entire YouTube is down now. So the video will be back (hopefully)...


Niki

Nah, it's up for me just fine,  I've also checked him out on reddit I first saw him online there, and he had some posts there under snicksound and it too seems 'gone'  he may have deleted all social media for some reason??  the 'deleted' account had links to snicksound.com which also is now a deadend domain.

~Phil

[synchu]

Sure. YouTube was down (https://www.theverge.com/2018/10/16/17987280/youtube-down-outage) worldwide and I thought this might be the reason of the video not being available.


And it happens at the same time I am looking for reactive load attenuators schematics and pointed to the good ol' EL34World - go figure :)


Anyone else tried this one, though?


Niki

[Willatr]

Interesting stuff !  I have a circuit question. I’m fairly new to electronic building but did successfully just build a simple tube amp so know enough to be dangerous.
What value of resistors are used, does the notation 2R for example mean pick a value R for the circuit and choose a resistor value 2xR or 8xR depending on the position or does it mean something different ? 

[brewdude]

Often “R” means ohms.
2R = 2 ohms

[Willatr]

Ok thanks ! This is my next project.

[silverfox]

Can't say for sure why, but I don't like the circuit as drawn. I believe it will attenuate the sound making it mushy. Call me a fuzzy mathematician . . .


Bit the topic made me wonder how a tube emulator might sound, like the old school tube EQ circuits. A reactive tube emulation circuit? Tube OP Amp?



silverfox.

[Willatr]

A couple more novice questions:-
Is the effective impedance ok for 8 ohm applications or 16 ohm ? If I need to, how do I modify for 8 ohm ?
What is the significance / impact of the inductor wire gage - does it play into the resonant frequency calculation ?
Do I need to be consider any interaction between the inductors due to physical location ?
Thanks

[PRR]

> impact of the inductor wire gage

Much larger is too expensive.

Much smaller is more loss (which may not be bad here).

The gauges listed "seem" to be taken from common hi-fi loudspeaker cross-over chokes. Hi-fi coils normally should have low loss. In this loss-box we could maybe use thinner gauge (and smaller cores). But pre-made chokes are a ton easier than winding and measuring chokes by hand.
https://www.parts-express.com/erse-10mh-18-awg-i-core-inductor-crossover-coil--266-550
https://www.parts-express.com/dayton-audio-10mh-18-awg-i-core-inductor-crossover-coil--257-550

[SnickSound]

I figured I'd update this since I've been silent for almost 2 years now. I've updated the first post.

In short: don't build this, it's a flawed design.

I'll be attempting JohnH's design instead: http://www.marshallforum.com/threads/simple-attenuators-design-and-testing.98285/
There's a similar basic idea, but at least he realized that you'd need reactive elements both in series and parallel with the speaker. It lacks the resonance boost, relying on the speaker to do that on its own (which works, just less predictable response from amp to amp).

What happened to my prototype is that the 2nd time I came to use it live I got no sound. Later, as I was trying to fight a lack of low end from my amp I realize it was running on just one power tube (2XEL34 Marshallesque build), and had been for an unknown amount of time (even with a tube out it was still pretty loud). I have no clue whether that tube had failed before, or whether it happened due to my prototype attenuator, but I didn't want to risk it anymore so I shelved it hoping to get back to it later. I finally did after deciding that a cranked Plexi type is what I need for a new band, and that's when I realized what I did wrong. So I salvaged it for parts that I will reuse to build JohnH's design. The big fat inductor will go into a reactive load (aka no speaker output) for recording later.

I did also delete my Youtube channel a while ago, for unrelated reasons. I don't think I'll be going back on the platform as I have limited time for this now and I found that too many people there just want to get schems and layouts without doing any homework to understand what they're doing and why they're doing it. My 6V6 Plexi video series was a good example of this, the initial video where I presented the planned schematic had thousands of view, but the follow up videos where I went over the changes and what didn't work with the initial plan had almost no engagement, yet I kept receiving messages to post a layout for that original schematic. When I started doing videos, I just wanted to share ideas with other builders, inspire people and have good quality demos of DIY amps. I don't like sharing schems and layouts as a "paint by number" approach, I don't have the technical background necessary. I make mistakes sometimes and I wanted to share these mistakes as part of the process when doing one-offs, not have people try to replicate my mistakes because they didn't watch the follow-ups. I may still post demo videos of completed builds in the future for the community, but I will keep discussion of schems and layouts to forums such as this one, where more interactive discussion is encouraged.